Framing Analysis

Framing analysis studies how texts make an issue intelligible by emphasizing some aspects of it and guiding the reader toward certain responses. The method asks how a text defines a problem and steers interpretation. Whether the text is accurate is a separate question.

Where does framing analysis fit? It is common in media studies and political communication, and IR research uses it widely. If your question asks how an issue is presented in texts, start here. For a closer focus on language and power, see Discourse Analysis.

What Is Framing Analysis?

The standard definition comes from Entman (1993). To frame is to “select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation.”

The definition names four functions.

Function	What it does	Example
Problem definition	Identifies what the issue is	“Immigration is a security threat” vs. “Immigration is an economic opportunity”
Causal interpretation	Attributes cause or responsibility	“Government policy caused the crisis” vs. “Global forces caused the crisis”
Moral evaluation	Makes a normative judgment	“This policy is unjust” vs. “This policy is necessary”
Treatment recommendation	Suggests what should be done	“Borders should be tightened” vs. “Integration programs should be expanded”

A given frame may accomplish only one or two of these functions in a particular text, but the four help you articulate how it constructs the issue and offers the audience an interpretive lens. Simply counting topics cannot do that.

Key Concepts

Frame. A central organizing idea that gives an issue meaning by emphasizing and connecting some aspects of reality while downplaying others (Gamson & Modigliani, 1989).

Framing devices. The textual cues that signal a frame, such as metaphors, catchphrases, exemplars, visual images, and appeals to principle (Gamson & Modigliani, 1989). These are what you code.

Generic vs. issue-specific frames. Generic frames (conflict, human interest, economic consequences, responsibility, morality) recur across issues and allow comparison across contexts (Semetko & Valkenburg, 2000). Issue-specific frames are built for one topic from your data or prior literature.

Episodic vs. thematic framing. Episodic frames focus on specific events or individual cases. Thematic frames place an issue in a broader context with systemic causes (Iyengar, 1991). In news coverage, episodic framing tends to individualize responsibility while thematic framing points to structural explanations.

Frame building vs. frame setting. Frame building is how some frames emerge and come to dominate media or political discourse. Frame setting is how media frames influence audience perceptions (Scheufele, 1999 and de Vreese, 2012). Most theses identify frames in texts, the frame-building side, because measuring frame setting requires experimental or survey data.

Salience. The mechanism that makes some information more noticeable and memorable (Entman, 1993). A frame needs no fabricated information. It works by emphasizing some elements and downplaying others.

Frames differ from topics. A topic is what a text is about (climate change). A frame is how the text presents it (climate change as an economic burden, a moral imperative, a scientific controversy, or a security threat). One topic can carry several competing frames. Students often mix the two up.

When to Use It

Framing analysis fits when your question asks how an issue is presented in communication.

Media coverage. How does the press frame a political crisis, policy debate, or social issue? Do frames differ across outlets or over time?
Political communication. How do political actors frame issues to build support, and how do competing frames interact in public debate?
Policy discourse. How do official documents define policy problems and justify solutions?
Comparative studies. How does framing of the same issue differ across countries, media systems, or languages?

Choose another method if the project is really about audience effects (experiments or surveys), about power and identity (discourse analysis), or about word frequencies (content analysis or text-as-data).

Ask yourself

Does your research question ask how something is presented, beyond what is covered?
Can you identify a specific set of texts (news articles, speeches, policy documents) to analyze?
Do you want to compare how different sources, countries, or periods present the same issue?
Is the framing of this issue contested, with actors presenting it in meaningfully different ways?

If you answered yes to most of these, framing analysis is likely a good fit.

This page covers written texts such as news and policy documents. Framing analysis also works for visual and social media content, but the coding rules change, so discuss the adaptation with your supervisor before you build the codebook.

How to Apply It

Step 1: Define Your Approach

First decide whether your frames come from prior literature, from your data, or from both.

Deductive (theory-driven). You start with frames drawn from the literature, such as the five generic news frames of Semetko and Valkenburg (2000), and look for them in your data. A deductive design is more manageable for BA students and easier to replicate.

Inductive (data-driven). You develop frames from the data through repeated reading and coding. This is necessary when no established typology fits your issue or when you want issue-specific frames. It is more demanding and requires careful documentation of how you identified each frame. Matthes and Kohring (2008) offer a cluster-analytic procedure for identifying frames inductively.

Combined approach. Start with frames from the literature and leave room for new frames that emerge from the data.

BA vs. MA expectations. BA students are generally well served by a deductive design using established generic frames, or a small set of issue-specific frames from prior research. MA students are expected to engage more critically with the choice. They may need to develop their own frame typology, justify it methodologically, and weigh its limitations honestly.

Step 2: Build Your Corpus

Corpus construction is a methodological decision you must justify. Settle four points.

Source selection. Which outlets, speakers, or document types, and why these? Tie the choice to your research question (for example, outlets with different editorial positions).
Time period. What period does the analysis cover, and why? A specific event, a policy cycle, or a long-term trend?
Sampling. Will you analyze all relevant texts in the period or a sample? If a sample, how is it drawn?
Size. The corpus must be large enough to support your claims. There is no universal minimum. A BA thesis might analyze 50-100 articles and an MA thesis 100-200 or more, depending on the method.

Building a Corpus covers corpus size, source selection, search strategies, and data management in detail.

Step 3: Develop a Codebook

A codebook defines each frame and the evidence required to apply it, in enough detail that another researcher could replicate your analysis. It should include the following.

Frame definitions. A clear, specific definition of each frame.
Indicators. Several concrete textual markers per frame (keywords, phrases, arguments, metaphors, narrative structures).
Coding rules. Explicit instructions for ambiguous cases.
- What is the unit of analysis (full article, paragraph, sentence)?
- Can one text receive several frames? If so, do you code a dominant frame or all frames present?
- What is the threshold for assigning a frame? Must it be the central organizing idea, or can it be secondary?
Examples. At least one clear example of each frame from your data or from prior literature.

Pilot your codebook. Test it on a small sample (10-15 texts) before coding the full corpus. The pilot exposes vague definitions, overlapping categories, and unused frames. Revise the codebook, and record what you changed and why for your methods section.

Step 4: Code Your Material

Read each text in full, headline and body together (see full article vs. headline framing below).
Identify which frames are present using the indicators in your codebook.
Record each coding decision in a structured format (spreadsheet, coding software, or a consistent table).
Keep notes on difficult or ambiguous cases. They feed your methodology and limitations sections.

Handling multi-frame texts. Many texts contain more than one frame, so set a decision rule in advance.

Dominant frame approach. Assign each text its single most prominent frame. This simplifies analysis but loses the secondary frames.
All-frames approach. Code every frame present. This captures more complexity, but you must define frame presence (for example, a frame must appear in at least two paragraphs or be supported by at least two distinct indicators).
Hierarchical approach. Code a primary frame and one or more secondary frames. This keeps the analysis manageable while preserving the secondary frames.

State your rule in the methodology chapter and apply it consistently.

Intercoder reliability. As a sole coder you cannot calculate formal intercoder reliability, so acknowledge it as a limitation. You can partly address it by asking your supervisor or a peer to code a small subsample, documenting your coding decisions, and discussing borderline cases openly in the analysis.

Step 5: Analyze and Interpret

Turn the coding into analysis.

Frame distributions. Which frames are most and least common? Report frequencies and percentages in tables or charts.
Patterns across sources. Do different outlets, speakers, or document types favor different frames?
Patterns over time. Do frames shift during your period? Are there moments where framing changes?
How each frame works. Give representative examples and show the language, arguments, and recurring narrative structure that carry each frame. Numbers alone do not show this.
Absent frames. Which perspectives are missing? If a frame the literature led you to expect is absent or marginal, that is a finding worth discussing.

Exercise: Draft your codebook

Develop a preliminary codebook for your framing analysis. For each frame, write:

A name and one-sentence definition
At least three textual indicators (keywords, phrases, argument types, or narrative patterns)
One example (real or hypothetical) showing how the frame would appear in a text from your corpus

Then write a short paragraph (150-200 words) explaining your coding rules. What is your unit of analysis? How will you handle texts that contain multiple frames? What is the threshold for assigning a frame?

Bring this codebook to your next supervision meeting.

Full Article vs. Headline Framing

Headlines are written to attract attention, often by editors, and may stress conflict that the article body does not sustain (or sit neutrally above a strongly framed article). The choice changes what you can claim.

Headlines only suit questions about first-glance presentation or how editors package stories, and allow a much larger sample.
Full articles capture the complete framing, including the causal attributions and remedies that usually appear in the body. They are feasible for a corpus under about 150 articles (BA) or about 200-250 (MA).
Both suit designs that compare headline and body framing, which can be a research question in itself.

Practical recommendation. For most theses, analyze full articles. If your corpus is too large, narrow it (fewer articles or a shorter period) instead of switching to headlines. If you do analyze headlines, state what that choice captures and what it misses.

Structuring Your Thesis

A framing thesis follows the standard structure. The method-specific content is as follows.

Introduction

Present the issue, explain why its framing matters, and state your research question.

Literature Review

Cover the substantive debate as well as framing theory. Say what is known about how this issue is framed and what remains unresolved. Introduce any frames you take from prior research and where they come from.

Analytical Framework / Methodology

Explain every decision well enough for another researcher to replicate the analysis.

Approach. Deductive, inductive, or combined, and why
Corpus. Source selection, time period, sampling strategy, and corpus size, with the reason for each decision
Codebook. Frames with definitions, indicators, examples, and decision rules. Say where deductive frames come from, or how inductive frames were developed
Coding procedure. Unit of analysis, multi-frame rule, and any reliability measures
Limitations. Single coder, language constraints, corpus boundaries

Findings

Organize the findings by frame (frequency, where it appears, how it works, how it relates to the others), by source or outlet for comparative designs, or by time period for crises, campaigns, and policy processes. Pair tables or charts of the distribution with excerpts that show how each frame operates.

Conclusion

Answer the research question and explain what the framing patterns mean for how the issue is constructed in public discourse, beyond summarizing results. Say whether your findings confirm, extend, or challenge prior research, and acknowledge limitations.

Example from the Literature

Semetko and Valkenburg (2000) is the standard model for deductive framing analysis. They analyzed 2,601 newspaper stories and 1,522 television news stories on European politics around the 1997 Amsterdam meeting of European heads of state, testing for five generic news frames. Each frame was operationalized as a set of yes/no questions (for the conflict frame, “Does the story reflect disagreement between parties/individuals/groups/countries?”) applied to every item in the corpus.

Irrespective of media type, responsibility and conflict were the most common frames and morality the least common. Frames did differ by outlet type, though, with responsibility used more in the serious press and human interest more in the sensationalist press. Two aspects of the study are worth emulating, a justified, deductively developed set of frames applied to a corpus large enough for meaningful comparison between outlet types, and specific indicators for every frame.

Common Pitfalls

1. Confusing frames with topics. This is the most common mistake (see Key Concepts). If your “frames” could serve as subject headings in a library catalog, they are probably topics.

2. Defining frames too vaguely. A frame that almost any text could be coded under is not analytically useful. Ask whether someone unfamiliar with your research could apply your codebook and reach similar decisions.

3. Inconsistent coding. Coding generously at the start and strictly at the end, or shifting your reading of a frame mid-corpus, undermines the results. Pilot the codebook, code in a consistent (or random) order, and revisit early decisions when you finish.

4. Presenting only numbers. “The conflict frame appeared in 43% of articles” is a starting point. Show how the frame operates in your texts. These examples demonstrate that your coding is valid and that the frames are real patterns rather than artifacts of your categories.

5. Skipping the design decisions. Absent frames are findings (Step 5). Corpus boundaries need a justification (Step 2), or readers will ask whether other choices would change the results. Texts often carry several frames, so a forced single-frame scheme needs a rule set in advance (Step 4).

Key Readings

How to use these readings. At minimum, read Entman (1993) and one methodology piece (Semetko & Valkenburg for deductive work, Matthes & Kohring for inductive work). Then look for framing studies on your topic. They show how the method has been applied in your area and may provide the frames you build on.

Discourse Analysis. Choose this when the project is less about coding frames and more about how language builds identities, authority, or power relations.
Building a Corpus. Use this before fixing the codebook if the source base is still unsettled. Frame counts mean little if the corpus boundaries are vague.